基于BP网络的谐波干扰误差修正方法在介质损耗角测量中的应用研究

Harmonic Interference Error Correction Method Based on BP Neural Network for Dielectric Loss Angle Measurement

摘要: 介质损耗角计算精度是影响氧化锌避雷器绝缘在线监测系统性能的关键因素，本文针对谐波对氧化锌避雷器绝缘在线监测系统中的介质损耗角计算精度的影响，提出了一种基于BP网络的谐波干扰误差修正方法，即用BP网络来辨识电力系统中的三次、五次谐波含量与介质损耗角误差修正值Δδ之间的非线性关系，构建基于BP网络的介质损耗角修正模型，并通过该修正模型得出介质损耗角修正值Δδ，来对在线监测系统介质损耗角测量值进行修正，仿真研究和实验室试验结果表明，该方法对于抑制谐波干扰误差具有良好的效果。

Abstract: The measurement precision of dielectric loss angle is a key factor of the performance of on-line monitoring system of metal-oxide arresters. The causes of data error of on-line insulating monitoring system for MOA are intro- duced in the paper. Questions about harmonic interference in on-line monitoring insulation system of MOA are stated. For the effect of harmonic interference in on-line monitoring insulation system, a harmonic interference error correction method based on BP neural network is presented for modifying dielectric loss angle. This method is used to identify the nonlinear relation between 3nd, 5th harmonic and dielectric loss angle modified value Δδ through BP network. Finally the adjusting dielectric loss angle model based on BP network is gained. The dielectric loss angle modified value Δδ, which can be got from the adjusting model, is used to modify the measured value of dielectric loss angle. The results from tests and simulation show that the two methods above-mentioned have a good result on eliminating harmonic interference error.